Peter Knížat, Statistical Office of the Slovak Republic, Slovak Republic

Type of article: scientific article
Pages: 21 – 38

Abstract

Web scraping provides a new innovative data source that can be utilised in price statistics by National Statistical Institutes. The prices of product-offers are automatically downloaded from the internet, which allows a wider selection of representative products in the consumer basket. The other advantage of scraping online prices is that the user can obtain characteristic parameters of individual products. These parameters are used in calculating a hedonic price index that is more preferable for products with a high replacement rate. The hedonic regression assumes that the natural logarithm of the product’s price can be explained by its characteristic parameters. In many previous researches, the authors estimate hedonic price indices for various products without any statistical verification of the fitted regression model.
In this paper, we carry out a thorough analysis of the hedonic regression model that encompasses checking the model’s diagnostics and its initial assumptions. Moreover, we use the analysis of variance to test contrasts between categories of individual characteristic parameters. The categories without any contrast are merged together and the hedonic price index is re-estimated. In the empirical study, we estimate and compare the hedonic price indices for different scenarios using observed web scraped prices from the Slovak market.